The Last Look…
Posted by Colin Lambert. Last updated: February 7, 2025
Sorry everyone but this is going to be an old-school rant – I try very hard not to compare today’s market to that of my day, but something today got me hot under the collar.
Apparently, according to three reports I read, dealers were “struggling” to cope with the weight of news as prices swung backwards and forwards on the tariff issue. I should stress these were well-respected outlets or analysts reporting this, not some retail, institutional-wannabee, trying to sound insightful (or more likely providing a rationale for all those margin calls).
It was a busy day, no doubt, and the reality of the Don’s second coming is finally hitting home – like the first, things will be chaotic, seemingly random, and liable to change on a minute-to-minute basis. This is what, traditionally, FX markets have thrived upon, so why are people “struggling”? Especially when they are meant to have technology at their fingertips to help them keep on top of the situation?
There are a few suggested themes from this, here are just three:
- Very few have really deployed AI to help analyse news, or if they have, it has failed miserably (it is reactive after all)
- Dealers are still scared to take a loss
- The business model is wrong in places, thanks to the absence of genuine risk appetite
The AI issue shouldn’t be a major surprise, it is still early days for people deploying the technology and it will always have limitations, in as much as it will always have limits. I recall being in a great presentation several years ago (my apologies I can’t remember the presenter but think he was from SunGard) who demonstrated live on stage, how easy it was to fool AI. A Boeing 747 identified as a donkey by the AI was my favourite – all through the addition of a small QR code in the corner of the picture.
Also, does AI only take the latest information, or does it try to build up a picture? As a dealer, you see the tariff issue as a binary one – they’re on, buy dollars (at least initially), they’re off, sell them. For AI perhaps, it’s a little more complicated, because it would only be a day or so since it learnt that tariffs were “certain”.
Speaking to a couple of people, I understand that the odd pricing engine struggled, spreads were as volatile as the market according to one buy-sider, and when that was the case, there was no human back-up. We can debate the rights and wrongs of asking for a price in a lump in such conditions, but another on the buy side told me they had to resort to the phone to get a larger ticket done, because the spreads were “unrealistic”.
It’s not the people, it’s the environment in which they have to operate
I am not sure I agree with this (without knowing the spread), because spreads are a good risk management tool in volatile markets, it’s just too many customers don’t understand that – they still think markets are back 2017, zero-interest bound, with next to no spreads in FX.
News alert: they’re not!
This buy-sider’s problem was not so much going to the phone – they like to keep their hand in I suspect – but more how long it took the LP to actually deliver a price (inevitably, an algo was offered first!) which they blamed on a lack of skill at the dealing desk. Again, I probably disagree (without knowing the circumstances), which brings me to the main point – it’s not the people, it’s the environment in which they have to operate.
When the smelly stuff hits the fan, dealers need leeway and risk limits, and it seems that too few actually have them. At one end of the scale, there are “LPs” who don’t want to quote anything over a million (They’re happy for you to do your 100 million in ones though funnily enough…), and at the other there are dealers who would love to take on some of this, but can’t because either risk management is the domain of the machine, or they simply don’t have the limits.
If things carry on as they have started – and there’s little reason to see why they shouldn’t – then I sense that a few businesses will regret not having a solid, human, risk function in FX. Volatile conditions do create challenges, but they also represent an opportunity, there are plenty of outlier trades by clients who may have either got their timing wrong, or don’t care about it at all – are these being exploited? I was speaking to a manual trader earlier, who was hugely frustrated because they saw positions coming in that they wanted to get on their book, but were thwarted because the machine internalised it before they could. It’s one thing hitting the internal machine, which can be done by some, but the skill of the trader is, as Kenny Rodgers says, knowing when to hold and when to fold – when things are going crazy, as they did for a period on Monday, it’s a lot harder under the current model, to actually grab the risk you want.
If chaotic conditions do continue into the year, I wonder whether there will be a re-think?
I should stress, I am not advocating a return to manual trading, or suggesting that things were better in my day, merely I am wondering if the pendulum has swung too far and we have forgotten the value of someone who will actually take on a position and not mind if things go a bit pear-shaped early? The problem with the “broker” model operated by so many LPs is that it doesn’t embrace three things: traders can’t cherry-pick the flow; there is no room for risk over anything other than a very short-term time horizon; and losses.
Unfortunately, when markets get crazy, things will go wrong, it’s natural. What worries me is too many institutions seem only to have losses in either extreme events like a flash crash, or when there is a technology failure. What happened to a dealer actually getting it wrong? As I like to say on the ACI Australia Dealing Course, if you meet a trader who has never lost money, you’ve met one of two things: A liar or a latency arbitrageur!
If chaotic conditions do continue into the year, I wonder whether there will be a re-think? I doubt it, to be honest, because by the time the decision is made, it will probably be 2026 and, history suggests, just in time for things to quiet down.
Ultimately, I don’t think this is a huge issue for the LPs – the increased volume still brings increased profits normally – it’s just quite a bit is left on the table – and increased income if customers are using algos. The platforms, as we have noted, have had a great start to the year (and I am reliably informed “enjoyed” Monday 3 February), so they too will thrive in these conditions.
This means the only ongoing issue from the chaos will be me getting irritated at the poor dealers “struggling”. Recently a discussion with friends on a WhatsApp group turned to our “triggers”. Of the four of us, they have one, maybe two each – we stopped counting mine at 25! I suspect we can make that 26…